CN109856080B - Near-infrared multispectral imaging multi-index collaborative nondestructive evaluation method for freshness of fish fillet - Google Patents

Abstract

The invention discloses a multi-spectral and multi-index synergistic fish fillet freshness evaluation method. The method respectively measures the freshness indexes TVB-N value, TBARS value and K value of fish fillet samples refrigerated for different days; uses a multi-spectral imaging system to scan corresponding The corresponding multi-spectral images were obtained from the fish fillet samples, and the near-infrared multi-spectral images were processed to extract the corresponding average reflectance spectral values at 5 central wavelengths of 1250 nm, 1452 nm, 1655 nm, 1785 nm, and 1890 nm; based on the obtained TVB ‑N value, TBARS value, K value and average spectral value, use LS‑SVM to build a prediction model, and predict the freshness of the fish fillet sample to be tested. The invention adopts multi-spectral and multi-index to synergistically evaluate the freshness of fish fillets, reduces the time required by the traditional method, enhances the detection efficiency and accuracy, and can effectively achieve the purpose of rapid, non-destructive and non-contact online detection.

Description

A non-destructive evaluation method for fish fillet freshness based on near-infrared multispectral imaging and multi-index synergy

technical field

The invention relates to the field of fish fillet freshness quality detection, in particular to a near-infrared multispectral imaging multi-index coordinated non-destructive evaluation method for fish fillet freshness.

Background technique

Fish is an important part of aquatic products. Fish is delicious and high in nutrients. It is an important source of protein, amino acids, fats and other nutrients needed by human beings. It is an important part of people's diet. Freshness is an important comprehensive index for fish quality evaluation. There are many factors that affect the freshness of fish, mainly related to storage temperature, microbial contamination, processing methods, and its own physicochemical and biochemical changes.

At present, the methods of measuring and evaluating the freshness of fish meat are roughly divided into: sensory evaluation method, physical property measurement, chemical analysis method, etc. The chemical analysis method commonly used in the laboratory is to measure the protein degradation index - volatile base nitrogen value (TVB-N), fat oxidation index - thiobarbituric acid value (TBARS) and ATP degradation index - K value to evaluate fish meat. Freshness. Generally speaking, when TVB-N value ≤ 15mg N/100g is defined as first-level freshness, 15mg N/100g < TVB-N value ≤ 20mg N/100g, it is defined as second-level freshness, TVB-N value > 20mg N /100g, it is defined as losing its edible value; similarly, when the K value is less than or equal to 20%, the fish is judged to be the first-grade freshness; when 20%<K value ≤60%, the fish is judged to be the second-grade freshness and can still be eaten. ; When the K value is greater than 60%, the fish meat has been spoiled and lost its edible value. In the laboratory, semi-micro nitrogen determination, spectrophotometry and high performance liquid chromatography are usually used to measure the corresponding three indicators. Chemical analysis tests, while accurate, are destructive tests. Obviously, in the actual detection process, these methods have cumbersome steps, high operation requirements, time-consuming and labor-intensive, and cannot achieve non-destructive and rapid online detection.

SUMMARY OF THE INVENTION

In order to overcome the above-mentioned shortcomings and deficiencies of the prior art, the purpose of the present invention is to provide a non-destructive evaluation method for fish fillet freshness with near-infrared multispectral imaging and multi-index synergy, which can effectively save the detection of fish fillets without destroying the fish fillet samples. Time, save measurement costs, and achieve rapid non-destructive testing and evaluation of fish fillet freshness.

The object of the present invention is achieved through the following technical solutions:

The non-destructive evaluation method of fish fillet freshness based on near-infrared multispectral imaging and multi-index synergy includes the following steps:

(1) Starting from day 0, with an interval of 1 or 2 days, prepare fish samples and refrigerate them for different days. The longest refrigerated days are no more than 7 days, and obtain N fish fillet samples, and the N fish fillet samples are randomly divided into M groups ; N is greater than 90, M is an integer from 4 to 10; the number of samples in each group is N/5;

(2) Using the near-infrared multispectral imaging system to scan the fish fillet samples of different storage days, and obtain a total of N multispectral images of the fish fillet samples;

(3) extracting the reflectance spectrum value corresponding to the multi-spectral central wavelength of the fish fillet sample, the central wavelengths are respectively 1250nm, 1452nm, 1655nm, 1785nm, and 1890nm;

(4) Determination of three indicators characterizing the freshness of fish fillets, using semi-micro nitrogen determination method to determine TVB-N value, spectrophotometric method to determine TBARS value and high performance liquid chromatography to determine K value;

(5) Combining the reflectance spectrum value corresponding to the central wavelength obtained in step (3) and the three fish fillet freshness index values of TVB-N value, TBARS value and K value obtained in step (4), use the least squares support vector Machine (LS-SVM) to build a multi-index prediction model of fish fillet freshness; Y _i =C ₀ +AX _1250nm +BX _1452nm +CX _1655nm +DX _1785nm +EX _1890nm ;

Among them, Y _i is the freshness evaluation index TVB-N value, TBARS value and K value, i is the freshness level, the values are 1, 2, and 0, respectively, representing the first-level freshness, the second-level freshness and the no freshness. Freshness; X _1250nm , X _1452nm , X _1655nm , X _1785nm , X _1890nm are the average reflection spectral values corresponding to the wavelengths of 1250nm, 1452nm, 1655nm, 1785nm, 1890nm, and are measured with TVB-N value, TBARS value and K value Corresponding to the spectral value at time; C ₀ , A, B, C, D, E are coordination coefficients, which are automatically generated by Matlab programming;

(6) Using the prediction model obtained in step (5) to evaluate the freshness of the fish fillet sample to be tested.

For further realizing the object of the present invention, preferably, the evaluation of the freshness described in step (6) is:

When the fish fillet is at the first level of freshness, the coordination coefficients of model Yi are C ₀ = _-22.31 , A=25.23, B=-21.42, C=46.55, D=124.12, E=23.48; the three measured simultaneously The variation ranges of the indicators are: TVB-N value≤14.27mg N/100g, TBARS value≤0.58mg/kg, K value≤19.36%;

When the fish fillet is at the second level of freshness, the coordination coefficients of model Yi are C ₀ = _-103.77 , A = 35.64, B = 41.72, C = 32.11, D = 165.69, E = 221.53; the three indicators measured at the same time The variation ranges of 14.27mgN/100g<TVB-N value≤19.88mgN/100g, 0.58mg/kg<TBARS value≤0.99mg/kg, 19.36%<K value≤59.48%;

When the fish fillets are not fresh, the coordination coefficients of model Yi are C ₀ = _-202.8 , A = 32.37, B = 46.42, C = 41.7, D = 195.13, e = 213.8; The variation ranges were: TVB-N value>19.88mgN/100g, TBARS value>0.99mg/kg, K value>59.48%.

Preferably, in step (3), the extraction of the reflectance spectral value corresponding to the multispectral central wavelength of the fish fillet sample is performed after performing size correction, masking and denoising processing on the obtained multispectral image of the fish fillet sample.

Preferably, the fish in the fish meat sample in step (1) is grass carp, common carp, silver carp, bullhead or herring.

Preferably, the preparation of the fish meat sample in step (1) includes descales, viscera, head, tail and skin, and is divided into 3cm×3cm×1cm in size; rinsed with running water, and blotted dry with absorbent paper The residual water on the surface of the fish was sealed in a polyethylene fresh-keeping bag and refrigerated at 4°C.

Preferably, in the M groups described in step (1), the number of samples in each group is the same or differs by one.

Preferably, M described in step (1) is 5.

The present invention loses edible value when the fish fillets are not fresh.

Compared with the prior art, the present invention has the following advantages and beneficial effects:

1) The present invention realizes rapid non-destructive testing and evaluation of the freshness of fish fillets on the premise of not destroying fish meat samples. Compared with conventional evaluation methods, the present invention is easy to operate, fast, non-destructive, non-contact, and does not require pretreatment of the samples. , which can realize fast and non-destructive online monitoring of fish fillet freshness.

2) The prediction model established by the present invention simultaneously measures and analyzes the three index values of the freshness, so that the evaluation of the freshness is more accurate and the error is smaller, which has a direct reality in order to ensure the quality and safety of fish meat and maintain the health of consumers. significance.

Description of drawings

FIG. 1 is a flow chart of a method for non-destructive evaluation of fish fillet freshness based on near-infrared multispectral imaging and multi-index synergy of the present invention.

Detailed ways

In order to better understand the present invention, the present invention will be further described below with reference to the accompanying drawings and embodiments, but the embodiments of the present invention are not limited thereto.

Example

As shown in Figure 1, the near-infrared multispectral imaging and multi-index synergy nondestructive evaluation method for fish fillet freshness includes the following steps:

(1) Prepare grass carp fillet samples and refrigerate them to obtain fish fillet samples of different refrigerated days: 10 grass carp (about 2kg in mass) were killed, scaled, gutted, head, tail and skin removed, and then divided into sizes 200 similar fish fillet samples (3cm×3cm×1cm), rinsed with running water, dried the residual water on the surface of the fish fillets with absorbent paper, sealed in polyethylene fresh-keeping bags and refrigerated at 4°C 2, 4, and 6 days, divided into four groups, each group randomly selected 50 fish fillets as fish fillet samples;

(2) Using the near-infrared multispectral imaging system (DL-604M) to scan the fish fillet samples with different refrigeration time, and obtain the multispectral images of 200 fish fillet samples;

(3) Perform size correction, masking, and denoising processing on the obtained multispectral images of 200 fish fillet samples, and extract the corresponding average reflection spectral values at 5 central wavelengths of 1250 nm, 1452 nm, 1655 nm, 1785 nm, and 1890 nm respectively;

(4) For 200 fish fillet samples, the semi-micro nitrogen determination method was used to determine the TVB-N value, the spectrophotometry method was used to determine the TBARS value, and the high performance liquid chromatography method was used to determine the K value. The linear transformation method was used to determine the three index values. After normalization, the measurement results are shown in Table 1;

Table 1 TVB-N value, TBA value and K value test results table

(5) Combine the average reflection spectrum value corresponding to the central wavelength obtained in step (3) and the normalized value of the three indicators obtained in step (4), use least squares support vector machine (LS-SVM) to construct fish fillets Freshness multi-index prediction model:

Y _i =C ₀ +AX _1250nm +BX _1452nm +CX _1655nm +DX _1785nm +EX _1890nm ;

Among them, Y _i is the freshness evaluation index TVB-N value, TBARS value and K value, i is the freshness level, the values are 1, 2, and 0, respectively, representing the first-level freshness, the second-level freshness and the no freshness. Freshness; X _1250nm , X _1452nm , X _1655nm , X _1785nm , X _1890nm are the average reflection spectral values corresponding to the wavelengths of 1250nm, 1452nm, 1655nm, 1785nm, 1890nm, and are measured with TVB-N value, TBARS value and K value Corresponding to the spectral value at time; C ₀ , A, B, C, D, E are coordination coefficients, which are automatically generated by Matlab programming;

When the fish fillet is at the first level of freshness, the coordination coefficients of model Yi are C ₀ = _-22.31 , A=25.23, B=-21.42, C=46.55, D=124.12, E=23.48; the three measured simultaneously The variation ranges of the indicators are: TVB-N value≤14.27mg N/100g, TBARS value≤0.58mg/kg, K value≤19.36%;

When the fish fillet is at the second level of freshness, the coordination coefficients of model Yi are C ₀ = _-103.77 , A = 35.64, B = 41.72, C = 32.11, D = 165.69, E = 221.53; the three indicators measured at the same time The variation ranges of 14.27mgN/100g<TVB-N value≤19.88mgN/100g, 0.58mg/kg<TBARS value≤0.99mg/kg, 19.36%<K value≤59.48%;

When the fish fillet is not fresh (loses its edible value), the coordination coefficients of model Yi are C ₀ = _-202.8 , A = 32.37, B = 46.42, C = 41.7, D = 195.13, e = 213.8; The variation ranges of the three indicators are: TVB-N value>19.88mg N/100g, TBARS value>0.99mg/kg, K value>59.48%.

(6) Using the prediction model obtained in step (5) to evaluate the freshness of the fish fillet sample to be tested.

In this example, the TVB-N value, TBARS value, and K value of the grass carp fillet samples that were refrigerated for 2 days and obtained by the constructed model were predicted by semi-micro nitrogen determination method using traditional methods to determine TVB-N value and spectrophotometric method. The determination of TBARS value and the determination of K value by high performance liquid chromatography are shown in Table 2. The fish fillets are in the first-level freshness level, and the experimental data obtained by the two methods are indistinguishable. The new method of the invention can be used instead of the traditional method to evaluate the fish. The freshness of the fish fillet has no difference in the measurement of the three indicators. These three indicators can be used to comprehensively evaluate the freshness of the fish fillet at the same time, which is more accurate and reliable than the evaluation of a single indicator.

The comparison of the new method invented in the embodiment of table 2 and the measurement value of the traditional measurement method (refrigerated for 2 days)

Example 2

The non-destructive evaluation method of fish fillet freshness based on near-infrared multispectral imaging and multi-index synergy includes the following steps:

(1) Prepare and refrigerate large head fish fillet samples to obtain fish fillet samples of different refrigeration days: 8 large head fish (about 2 kg in mass) are killed, descaled, gutted, head, tail and skin removed, and then divided into sizes 160 fish fillet samples of similar size (3cm×3cm×1cm), rinsed with running water, dried the residual water on the surface of the fish fillets with absorbent paper, sealed in polyethylene fresh-keeping bags and refrigerated at 4°C. , 1, 3, and 5 days, divided into four groups, each group randomly selected 40 fish fillets as fish fillet samples;

(2) Using the near-infrared multispectral imaging system (DL-604M) to scan the fish fillet samples with different refrigeration time, and obtain the multispectral images of 160 fish fillet samples;

(3) Perform size correction, masking, and denoising on the multispectral images of the 160 fish fillet samples obtained, and extract the corresponding average reflection spectral values at 5 central wavelengths of 1250 nm, 1452 nm, 1655 nm, 1785 nm, and 1890 nm respectively;

(4) Use semi-micro nitrogen determination method to measure TVB-N value, spectrophotometric method to measure TBARS value and high performance liquid chromatography to measure K value, and normalize the three index values, the measurement results are shown in Table 3 Show;

Table 3 uses traditional method to measure TVB-N value, TBA value and K value results

(5) Combine the average reflection spectrum value corresponding to the central wavelength obtained in step (3) and the normalized value of the three indicators obtained in step (4), use least squares support vector machine (LS-SVM) to construct fish fillets Freshness multi-index prediction model;

The described prediction model of step (5), the model equation is specifically:

Y _i =C ₀ +AX _1250nm +BX _1452nm +CX _1655nm +DX _1785nm +EX _1890nm ;

Among them, Y _i is the freshness evaluation index TVB-N value, TBARS value and K value, i is the freshness level, the values are 1, 2, and 0, respectively, representing the first-level freshness, the second-level freshness and the no freshness. Freshness; X _1250nm , X _1452nm , X _1655nm , X _1785nm , X _1890nm are the average reflection spectral values corresponding to the wavelengths of 1250nm, 1452nm, 1655nm, 1785nm, 1890nm, and are measured with TVB-N value, TBARS value and K value Corresponding to the spectral value at time; C ₀ , A, B, C, D, E are coordination coefficients, which are automatically generated by Matlab programming;

When the fish fillet is at the first level of freshness, the coordination coefficients of model Yi are C ₀ = _-22.31 , A=25.23, B=-21.42, C=46.55, D=124.12, E=23.48; the three measured simultaneously The variation ranges of the indicators are: TVB-N value≤14.27mg N/100g, TBARS value≤0.58mg/kg, K value≤19.36%;

When the fish fillet is at the second level of freshness, the coordination coefficients of model Yi are C ₀ = _-103.77 , A = 35.64, B = 41.72, C = 32.11, D = 165.69, E = 221.53; the three indicators measured at the same time The variation ranges of 14.27mgN/100g<TVB-N value≤19.88mgN/100g, 0.58mg/kg<TBARS value≤0.99mg/kg, 19.36%<K value≤59.48%;

When the fish fillet is not fresh (loses its edible value), the coordination coefficients of model Yi are C ₀ = _-202.8 , A = 32.37, B = 46.42, C = 41.7, D = 195.13, e = 213.8; The variation ranges of the three indicators are: TVB-N value>19.88mg N/100g, TBARS value>0.99mg/kg, K value>59.48%.

(6) Using the prediction model obtained in step (5) to evaluate the freshness of the fish fillet sample to be tested.

In this example, the TVB-N value, TBARS value, and K value of the grass carp fillet samples that were refrigerated for 5 days and were predicted by the constructed model were compared with the TVB-N value measured by the semi-micro nitrogen determination method and the spectrophotometric method using the traditional method, respectively. The determination of TBARS value and the determination of K value by high performance liquid chromatography are shown in Table 4. The fish fillets are in the second-level freshness level, and the experimental data obtained by the two methods are indistinguishable. The new method of the invention can be used instead of the traditional method to evaluate the fish. The freshness of the fish fillet has no difference in the measurement of the three indicators. These three indicators can be used to comprehensively evaluate the freshness of the fish fillet at the same time, which is more accurate and reliable than the evaluation of a single indicator.

Comparison of the new method invented in the embodiment of table 4 and the measurement value of traditional measurement method (refrigerated for 5 days)

The embodiments of the present invention are not limited by the examples, and any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention should be equivalent substitution methods. Included in the protection scope of the present invention.

Claims (4)

1. The nondestructive evaluation method for the freshness of the fillets with near-infrared multispectral imaging and multi-index cooperation is characterized by comprising the following steps of:

(1) from 0 day, every 1 or 2 days, preparing fish samples and refrigerating for different days, wherein the longest refrigerating day is not more than 7 days, and obtaining N fish slice samples which are randomly divided into 5 groups; n is more than 90, and the number of samples in each group is N/5;

(2) scanning the fillet samples with different storage days by using a near-infrared multispectral imaging system to obtain multispectral images of N fillet samples;

(3) extracting reflection spectrum values corresponding to multispectral central wavelengths of the fish slice sample, wherein the central wavelengths are 1250nm, 1452nm, 1655nm, 1785nm and 1890nm respectively;

(4) measuring three indexes representing the freshness of the fish fillets, measuring a TVB-N value by using a semi-micro azotometry, measuring a TBARS value by using a spectrophotometry and measuring a K value by using a high performance liquid chromatography;

(5) combining the reflection spectrum value corresponding to the central wavelength obtained in the step (3) and three fillet freshness index values of the TVB-N value, the TBARS value and the K value obtained in the step (4), and constructing a fillet freshness multi-index prediction model by using a least square support vector machine; y is_i=C₀+AX_1250nm+BX_1452nm+CX_1655nm+DX_1785nm+EX_1890nm；

Wherein, Y_iThe index TVB-N value, TBARS value and K value are freshness evaluation indexes, i is a freshness grade, the values are 1, 2 and 0 respectively, and the first-level freshness, the second-level freshness and the no-freshness are respectively represented; x_1250nm、X_1452nm、X_1655nm、X_1785nm、X_1890nmAverage reflectance spectrum values corresponding to wavelengths of 1250nm, 1452nm, 1655nm, 1785nm and 1890nm respectively, and corresponding to spectrum values at the time of measurement of TVB-N value, TBARS value and K value; c₀A, B, C, D, E are coordination coefficients which are automatically generated through Matlab programming;

(6) evaluating the freshness degree of the fillet sample to be tested by using the prediction model obtained in the step (5);

the freshness was evaluated as:

when the fillets are in first-grade freshness, the model Y_iThe coordination coefficients are respectively C₀= 22.31, a =25.23, B = -21.42, C =46.55, D =124.12, E = 23.48; the variation range of three indexes measured simultaneouslyRespectively, the following steps: the TVB-N value is less than or equal to 14.27mgN/100g, the TBARS value is less than or equal to 0.58mg/kg, and the K value is less than or equal to 19.36 percent;

when the fillet is in the second-level freshness degree, the model Y_iThe coordination coefficients are respectively C₀= 103.77, a =35.64, B =41.72, C =32.11, D =165.69, E = 221.53; the variation ranges of the three indexes measured simultaneously are respectively as follows: the TVB-N value is more than 14.27mg N/100g and less than or equal to 19.88mg N/100g, the TBARS value is more than 0.58mg/kg and less than or equal to 0.99mg/kg, and the K value is more than 19.36 percent and less than or equal to 59.48 percent;

when the fillet is at non-freshness, the model Y_iThe coordination coefficients are respectively C₀= 202.8, a = 32.37, B =46.42, C =41.7, D =195.13, e = 213.8; the variation ranges of the three indexes measured simultaneously are respectively as follows: TVB-N value is more than 19.88mg N/100g, TBARS value is more than 0.99mg/kg, and K value is more than 59.48%.

2. The near-infrared multispectral imaging multiindex collaborative fillet freshness nondestructive evaluation method according to claim 1, wherein the extraction of the reflectance spectrum value corresponding to the multispectral center wavelength of the fillet sample in the step (3) is performed after size correction, masking and denoising of the obtained multispectral image of the fillet sample.

3. The near-infrared multispectral imaging multi-index collaborative fish slice freshness nondestructive evaluation method according to claim 1, wherein the fish of the fish sample in the step (1) is grass carp, silver carp, big head fish or black carp.

4. The near-infrared multispectral imaging multiindex coordinated fillet freshness nondestructive evaluation method according to claim 1, wherein the preparation of the fish meat sample in the step (1) comprises scaling, eviscerating, decapitating and skinning, and the fish meat sample is divided into 3cm x 1cm in size; washing with flowing water, sucking residual water on the surface of fish meat with absorbent paper, sealing in polyethylene freshness protection package, and refrigerating at 4 deg.C.

Images